<!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. -->
bert-base-cased-cola
This model is a fine-tuned version of bert-base-cased on the GLUE COLA dataset. It achieves the following results on the evaluation set:
- Loss: 0.3976
- Matthews Correlation: 0.5795
Model description
More information needed
Intended uses & limitations
More information needed
Training and evaluation data
More information needed
Training procedure
Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 2e-05
- train_batch_size: 16
- eval_batch_size: 8
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_ratio: 0.06
- num_epochs: 10.0
Training results
| Training Loss | Epoch | Step | Validation Loss | Matthews Correlation |
|---|---|---|---|---|
| 0.5733 | 0.93 | 500 | 0.4877 | 0.5001 |
| 0.3426 | 1.87 | 1000 | 0.3976 | 0.5795 |
| 0.2321 | 2.8 | 1500 | 0.6284 | 0.6034 |
| 0.1599 | 3.74 | 2000 | 0.6997 | 0.5718 |
| 0.123 | 4.67 | 2500 | 0.8962 | 0.5624 |
| 0.0888 | 5.61 | 3000 | 0.8743 | 0.5857 |
| 0.0659 | 6.54 | 3500 | 0.9159 | 0.5962 |
| 0.0385 | 7.48 | 4000 | 1.2283 | 0.5704 |
| 0.0299 | 8.41 | 4500 | 1.2601 | 0.5662 |
| 0.0211 | 9.35 | 5000 | 1.2885 | 0.5629 |
Framework versions
- Transformers 4.21.3
- Pytorch 1.7.1
- Datasets 1.18.3
- Tokenizers 0.11.6
